Nvidia's $250 billion OpenAI backstop + ChatGPT's bioweapon problem

The Wall Street Journal reports that Nvidia is in talks to guarantee $250 billion in financing so OpenAI can lease a 10-gigawatt campus that SoftBank's energy arm is developing in Piketon, Ohio, on the site of a former uranium enrichment plant. The full project could cost more than $500 billion, and Nvidia is separately discussing another $350 billion tied to chip purchases. Investor Michael Burry, who is short Nvidia, called the structure circular — a chip vendor guaranteeing its own customer's spending on its own chips.

The Wall Street Journal reports that hundreds of people have asked ChatGPT for help making poisons and biological weapons since last summer, and that some received step-by-step instructions OpenAI staff described as simple enough for a high-school biology student to follow. OpenAI internally flagged GPT-5 as high-risk for exactly this in mid-2025, then downgraded the rating that autumn. The company suspended the accounts but did not notify authorities, which it is not legally required to do.

CXMT, China's largest maker of dynamic random-access memory (DRAM), closed up 466% on its first day of trading in Shanghai, valuing the Hefei company at more than $487 billion and making it the most valuable firm listed on a mainland Chinese exchange. The offering raised at least $8.6 billion. First-quarter revenue rose more than 700% year-over-year on AI demand, and CXMT is China's best-positioned attempt to build high-bandwidth memory (HBM) at home while US export controls hold.

Nvidia and South Korea's SK Group announced an expanded partnership valued at over $500 billion, covering multi-gigawatt AI factories and a long-term memory supply agreement. SK Telecom will build a 2-gigawatt facility in Korea running Nvidia's next-generation Vera Rubin systems on SK hynix HBM4 memory, with the first phase due online in 2027. The memory half matters most: high-bandwidth memory has become the tightest constraint on shipping AI hardware.

Hugging Face chief executive Clem Delangue asked OpenAI to release the complete traces from the agents that broke into his company's systems, and to commit $100 million in compute so the open-source community can build defenses with it. "The first autonomous agent cyberattack is an unprecedented event. It deserves an unprecedented response," he wrote. OpenAI confirmed the two sides met, called the incident an important moment for AI safety, and said a technical report is coming in the next few weeks.

Bloomberg reports that Apple now aims to unveil its smart glasses, code-named N50, at its developer conference in June 2027 with a release late that year — roughly six months later than planned. The delay is partly about privacy: Apple is weighing whether to ship cameras at all, or to allow them only as eyes for Siri rather than for user-initiated photos and video. Meta has sold more than seven million Ray-Ban glasses and absorbed most of the resulting backlash.

TechCrunch visited Encord's facility in San Leandro, California, where workers wearing head-mounted sensors from the German startup Zander Labs generate training data for robotic manipulation. The headsets read mental states like error, intent, and surprise, alongside first-person video and forearm muscle sensors. The premise is that physical AI is bottlenecked on data rather than on models — as Encord's head of robot learning put it, "The data simply does not exist."